Ideological education intelligent evaluation system based on artificial intelligence data fusion

Abstract

The invention relates to the technical field of education, and discloses an intelligent evaluation system for ideological education based on artificial intelligence data fusion, and the system comprises a data obtaining and processing module which is used for collecting online behavior data of students from a plurality of data sources; the dynamic thought cognition model module is used for processing modal student data; the model training and optimizing module is used for acquiring model data in the module; the evaluation and prediction module is used for estimating the thought cognition state of the student in real time by using the optimization model data; the attribution analysis module is used for utilizing an evaluation result and optimizing model data; the intelligent intervention path recommendation module is used for recommending the thought targets and evaluation results of the students. On-line behaviors, off-line performance and subjective feedback data are integrated by adopting a multi-modal data acquisition and processing technology, so that the technical effect of comprehensively and accurately analyzing the thought state of the student is achieved. According to the scheme, limitation caused by a single data source in a traditional method is eliminated, and multi-angle evaluation of student information is achieved.

Description

Technical Field

[0001] This invention relates to the field of educational technology, specifically to an intelligent assessment system for ideological education based on artificial intelligence data fusion. Background Technology

[0002] In the field of modern educational assessment, traditional methods face several significant shortcomings, especially in accurately understanding students' mental states. First, many existing technologies rely solely on test scores or classroom performance as the primary basis for assessment. This single source of data limits a comprehensive understanding of students' overall performance and often fails to reflect students' emotional changes and motivations during the learning process.

[0003] The use of static models is prevalent in current assessment systems. These models cannot track the dynamic changes in students' mental states in real time and lack flexibility. For example, when students face sudden stress or challenges, static models may not be able to adjust assessment criteria in a timely manner, making it difficult for educators to quickly develop appropriate intervention measures. This lag effect not only affects the effectiveness of intervention strategies but may also exacerbate students' mental distress.

[0004] Many educational intervention programs lack personalization and are implemented in a "one-size-fits-all" manner. This approach fails to provide effective support to students from diverse backgrounds and with varying needs when they receive the same interventions. For example, if a teaching strategy is based solely on general applicability, some individual students may not benefit at all, and in some cases, their frustration may even increase. Such interventions not only fail to improve learning outcomes but may also have negative effects.

[0005] The lack of utilization of emerging data types is also a shortcoming of existing technologies. With the development of technology, multimodal data such as online behavior and subjective feedback are becoming increasingly abundant, but traditional evaluation methods have failed to fully integrate this information, leading to biased evaluation results. At the same time, the complexity and diversity of data processing are not accurately reflected in existing technologies, hindering more accurate decision-making. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an intelligent assessment system for ideological education based on artificial intelligence data fusion, which solves the problems of incomplete assessment of students' ideological state, insufficient dynamic monitoring, and inadequate personalized intervention in traditional educational assessment methods.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an intelligent assessment system for ideological education based on artificial intelligence data fusion, comprising: The data acquisition and processing module is used to collect students' online behavior data, offline performance data and subjective feedback data from multiple data sources, and to clean, fuse, label and time-series align the data to generate multimodal student data. The dynamic thought and cognition model module is connected to the data acquisition and processing module. Through the processed modal student data, it adopts a dynamic state-space model and uses state transition equations and observation equations to describe the evolution of students' thought and cognition states, thereby generating model data. The model training and optimization module is connected to the dynamic thought cognition model module. By acquiring the model data in this module, it trains and optimizes the model parameters to generate optimized model data. The assessment and prediction module, which is connected to the model training and optimization module, uses the optimized model data to estimate the students' cognitive state in real time, and generates the future cognitive trajectory based on the model's prediction results, thus generating the assessment results. The attribution analysis module, which is connected to the assessment and prediction module and the model training and optimization module, uses the assessment results and optimized model data to quantitatively analyze the impact of different educational activities on changes in students' cognitive state and generate an attribution report with visualization effects. The intelligent intervention path recommendation module, which is connected to the assessment and prediction module and the attribution analysis module, plans the optimal educational intervention path using a cost function based on the student's ideological goals and assessment results.

[0008] Preferably, the data acquisition and processing module specifically includes: Online learning data automatically collected by the learning management system includes course access records, assignment submission status, and forum interaction data; The acquired text data is transformed into a high-dimensional semantic vector representation using a pre-trained language model to facilitate subsequent analysis.

[0009] Furthermore, the data acquisition and processing module collects students' online behavior data, offline performance data, and subjective feedback data from multiple data sources. By cleaning, fusing, labeling, and aligning this data temporally, the system generates high-quality multimodal student data. This process addresses the problems of single data sources and incomplete assessments in traditional methods, ensuring a multi-faceted and accurate evaluation.

[0010] Preferably, the state transition equation of the dynamic thought cognition model module is: ; In the formula, Represents an individual In time The hidden thought state vector, Represents a nonlinear function. Represents an individual In time The state transition matrix, This represents the hidden state vector from the previous time step. Indicates the input control matrix. Represents the external intervention vector. Indicates process noise; And the observation equation is: ; In the formula, Represents an individual In time Observational data, Indicates parameters The nonlinear observation function, Represents the observation matrix. This indicates observation noise.

[0011] Furthermore, the dynamic thought and cognition model module, by applying a dynamic state-space model and combining state transition equations and observation equations, enables real-time tracking of students' thought and cognition states at various time points. This innovative modeling method addresses the shortcomings of traditional static models in adapting to changes in students' states, allowing educators to make timely adjustments and decisions based on the latest data.

[0012] Preferably, the model training and optimization module uses a combination of variational autoencoder and expectation-maximization algorithm to train the optimized model data in order to maximize the accuracy of the model.

[0013] Furthermore, the model training and optimization module utilizes a combination of variational autoencoders and the expectation-maximization algorithm to train and optimize the model data. This approach not only effectively improves the model's accuracy but also endows it with stronger adaptability, enabling it to better handle various complex student data.

[0014] Preferably, the evaluation and prediction module includes the following steps: Receive model data and optimize model data; The extended Kalman filter algorithm is used to estimate the students' cognitive state in real time, and the state estimate is updated at each time point. Based on the current state estimate and external intervention information, the trajectory of thought state over a future period is generated through model prediction and calculation. The generated thought state trajectory and its changing trend are output as the evaluation result.

[0015] Preferably, the assessment and prediction module uses multiple assessment indicators to comprehensively analyze the assessment results and generate personalized assessment reports for different student groups, including the development trend of their ideological state and the gap with the target ideological state, so as to help educators formulate corresponding educational strategies.

[0016] Furthermore, the assessment and prediction module focuses on using the optimized data output by the model training and optimization module to estimate students' cognitive states in real time. Through the extended Kalman filter algorithm, this module can update the state estimates at various time points and generate a trajectory of students' cognitive states for a future period based on the current state estimate and external intervention information.

[0017] Preferably, the attribution analysis module includes the following steps: Obtain evaluation results and optimize model data; Multiple quantitative indicators are defined for each educational activity to quantify changes in students' mental state; Statistical analysis methods were used to analyze the quantitative indicators and determine the impact weight of each educational activity. Generate attribution reports to demonstrate the specific impact of each educational activity on changes in students' cognitive state, for educators' reference.

[0018] Preferably, the attribution analysis module uses the external intervention matrix and evaluation results to perform regression analysis to quantify the direct and indirect impacts of different educational activities on changes in students' mental state, and generates a visual report to show the effect analysis of each intervention measure.

[0019] Furthermore, the attribution analysis module quantitatively analyzes the evaluation results and optimized model data to demonstrate the impact of different educational activities on changes in students' cognitive states. This analysis not only facilitates the evaluation of educational activity effectiveness but also provides educators with visual reference reports to help them develop more targeted educational strategies.

[0020] Preferably, the intelligent intervention path recommendation module includes the following steps: Receive the assessment results and attribution reports; Based on the assessment results, set target mental states and analyze the key factors affecting mental states; The optimal intervention path from the current state of mind to the target state of mind is calculated using a cost function based on an optimization algorithm; It provides recommended interventions and corresponding implementation suggestions to support educators in developing personalized teaching plans.

[0021] Preferably, the intelligent intervention path recommendation module uses a cost function for optimization, which includes: ; In the formula, This represents the total cost of the entire path, starting from the current time. Until the end time Total loss, Indicates each time step The costs are accumulated over time. Start to Time Finish, Represents an individual In time The hidden thought state vector, It is the target thought state vector. Denotes the weighted square norm. Indicates time The applied intervention vector, The weighted square norm representing the intensity of the intervention. This indicates the degree of sensitivity to control costs.

[0022] Furthermore, the intelligent intervention path recommendation module comprehensively considers assessment results and attribution reports, and uses a cost function to plan the optimal educational intervention path based on students' ideological goals. The innovation of this module lies in providing personalized intervention suggestions through structured data and optimization algorithms, effectively improving the scientific rigor and relevance of educators' lesson planning.

[0023] This invention provides an intelligent assessment system for ideological education based on artificial intelligence data fusion. It has the following beneficial effects: 1. This invention integrates online behavior, offline performance, and subjective feedback data using multimodal data acquisition and processing technology, achieving a comprehensive and accurate analysis of students' mental state. This approach eliminates the limitations of single data sources in traditional methods, enabling multi-faceted evaluation of student information.

[0024] 2. This invention utilizes a dynamic cognitive model, combined with state transition equations and observation equations, to achieve real-time tracking and analysis of students' cognitive states. This dynamic modeling method effectively addresses the shortcomings of existing static assessment models, enabling educational interventions to be adjusted promptly based on the latest data.

[0025] 3. This invention utilizes intelligent intervention path recommendation technology to provide educators with personalized intervention strategies. This technical solution optimizes students' mental state and learning outcomes by planning the optimal path through a cost function. Compared to existing methods, this solution addresses the problem of insufficient precision in intervention measures, ensuring that educators can develop more effective personalized teaching plans. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the system framework of the present invention; Figure 2 This is a schematic diagram of the data acquisition and processing module architecture of the present invention; Figure 3 This is a schematic diagram of the dynamic thought cognition model module architecture of the present invention; Figure 4 This is a schematic diagram of the evaluation and prediction module architecture of the present invention; Figure 5 This is a schematic diagram of the attribution analysis module architecture of the present invention. Detailed Implementation

[0027] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Please see the appendix Figure 1 - Appendix Figure 5 This invention provides an intelligent assessment system for ideological education based on artificial intelligence data fusion, comprising: The data acquisition and processing module is used to collect students' online behavior data, offline performance data and subjective feedback data from multiple data sources, and to clean, fuse, label and time-series align the data to generate multimodal student data. Specifically, in this embodiment, the data acquisition and processing module efficiently and accurately collects and processes various types of student data to construct multimodal student data that can be used for subsequent analysis and modeling.

[0029] The data acquisition and processing module comprises several components, namely the data acquisition unit, data cleaning unit, data fusion unit, data labeling unit, and time sequence alignment unit. These components are closely connected to form a complete data processing flow.

[0030] The data collection department automatically collects students' online behavior data through the learning management system. Specifically, this includes course access records, assignment submissions, online test results, and forum interaction data. This online behavior data provides structured information on students' learning engagement, facilitating a comprehensive assessment of students' mental state.

[0031] The data cleaning department preprocesses the acquired raw data to remove redundant information and outliers. The cleaning steps include missing value handling, noise detection, and inconsistent data standardization. The output of this step is the processed and cleaned data, ensuring data quality and minimizing interference with subsequent analysis.

[0032] The Data Fusion Department integrates information from different data sources (online behavioral data, offline performance data, and subjective feedback data). The fusion process uses a weighted average method, where weights are set based on the importance and reliability of the data sources, as shown in the following formula: ; In the formula, This indicates the result of the data fusion. This indicates the number of data sources participating in the fusion. Indicates the first The original data from each data source, Indicates the first The weight of each data source, This indicates the accumulation from the first data source to the second. One data source.

[0033] The data labeling department annotates multimodal student data to facilitate feature extraction during subsequent model training. The labeling process is algorithm-based, including rule-based labeling and text classification techniques, ensuring that each record clearly identifies the type of information it contains (academic performance, interaction frequency, feedback sentiment, etc.). This process contributes to the establishment and optimization of subsequent analysis models.

[0034] The temporal alignment section ensures that all data is arranged according to a time series and is consistent across the time dimension. This process standardizes the time signatures of each data source by calculating timestamps, generating a unified time baseline to ensure the accuracy and consistency of subsequent dynamic models.

[0035] For each of the above parts, a data acquisition department is first established to obtain multimodal data. Then, the data is preprocessed by the data cleaning department, further processed by the data fusion department for weighted fusion, labeled by the data labeling department, and finally sorted by the time-series alignment department to output the generated multimodal student data.

[0036] The dynamic thought and cognition model module is connected to the data acquisition and processing module. Through the processed modal student data, it adopts a dynamic state-space model and uses state transition equations and observation equations to describe the evolution of students' thought and cognition states, thereby generating model data. Specifically, in this embodiment, the dynamic thought cognition model module accurately models the students' thought cognition state so as to make dynamic assessments based on real-time data changes.

[0037] The dynamic thought and cognition model module consists of two main parts: state transition equations and observation equations. Based on the state space model, it constructs a student thought state evolution model and combines teacher-student interaction and external intervention information to achieve timely and accurate assessment of students' thought and cognition states.

[0038] First, a state transition equation is set up to describe the individual's state transition over time. How does the hidden thought state evolve over time? The mathematical model of the state transition equation is as follows: ; In the formula, Represents an individual In time The hidden thought state vector, Represents a nonlinear function. Represents an individual In time The state transition matrix, This represents the hidden state vector from the previous time step. Indicates the input control matrix. Represents the external intervention vector. Indicates process noise; And the observation equation is: ; In the formula, Represents an individual In time Observational data, Indicates parameters The nonlinear observation function, Represents the observation matrix. This indicates observation noise.

[0039] To dynamically update the model, a state-space model is first established, defining the sample data and model parameters. Next, the model parameters are continuously optimized using feedback mechanisms and reinforcement learning methods. The specific implementation process includes the following steps: Initialization: Set the initial state Covariance Matrix The model is initialized using historical data.

[0040] State prediction: Based on the state transition equation, the current state is predicted, as shown in the following formula: ; In the formula, Indicates time Prediction of the current state Covariance Update: Calculate and update the state covariance using the following formula: ; In the formula, Indicates time At that time, the covariance prediction matrix of the state, Indicates the first An object in time The state transition matrix, Indicates time The posterior state covariance matrix at time , express The transpose of is used to propagate uncertainty. This represents the process noise covariance matrix.

[0041] Observation Update: The current observation data is fused using observation equations to correct the prediction status. ; In the formula, Represents the Kalman gain matrix. Indicates time Time based The predicted state covariance matrix, Represents the observation matrix. express transpose, The covariance matrix representing the observation noise. It represents the total uncertainty of the observation-prediction covariance plus the observation noise.

[0042] State Update: The state and covariance are finally updated using the following formula: ; In the formula, Represents an individual In time The updated state estimate, Represents an individual The predicted state, Indicates Kalman gain, Represents an individual In time The actual observed value, Represents the observation matrix. This represents a nonlinear observation function.

[0043] ; In the formula, This represents the updated state covariance matrix. Represents the identity matrix. This represents the gain portion used to correct for the predicted covariance. This represents the predicted covariance matrix.

[0044] Through the above steps, the model parameters are continuously iterated and updated, enabling the dynamic cognitive model to assess students' mental states in real time. The model outputs the hidden states. and observation data It can be used in the subsequent assessment and prediction module, which can implement subsequent personalized educational interventions based on the individual's dynamic state.

[0045] The model training and optimization module is connected to the dynamic cognitive model module. By acquiring model data from this module, it trains and optimizes model parameters to improve model accuracy and generate optimized model data. Specifically, in this embodiment, the model training and optimization module aims to train, optimize, and adjust the parameters of the model based on the output data of the dynamic cognitive model, thereby improving its prediction accuracy and adaptability.

[0046] The model training and optimization module comprises multiple functional components, including a data input section, a model training section, a parameter optimization section, and a validation and testing section. These components are closely related through data transmission to form a complete model training chain.

[0047] First, the data input unit acquires multimodal data from the dynamic thought cognition model module, including students' thought state vectors. and observed values This data, after preliminary processing, serves as input for the model training department. The main purpose of this step is to ensure the integrity and quality of the input data, thereby improving the effectiveness of subsequent modeling.

[0048] Next, the model training department will establish a training model through the following steps. This model can be a linear regression, decision tree, or deep learning model. Taking a deep neural network as an example, the model is defined as follows: ; In the formula, It is an activation function. It is a weight matrix. It is a bias term.

[0049] The relevant loss function is used to measure the difference between the predicted result and the actual observed value. The commonly used loss function is the mean squared error (MSE): ; Here, For loss function, Based on actual observation data, The total number of samples, The data predicted by the model. This is achieved by minimizing the loss function. This can optimize model parameters. and .

[0050] To optimize the parameters, the optimization department uses backpropagation combined with gradient descent to update the parameters. In each iteration, the gradient of each parameter is calculated based on the loss function, as shown in the formula: ; ; In the formula, Represents the model's weight parameters. These represent the bias parameters of the model. Indicates the learning rate. This indicates that the loss function is applied to the weights. The partial derivatives, The loss function is expressed as a function of bias. The partial derivatives of .

[0051] Subsequently, the validation and testing department evaluates the trained model. The model's performance on the validation set is assessed to confirm its generalization ability. The specific steps are as follows: the validation set data is input into the model to obtain predicted values, and then various performance metrics, such as accuracy and F1 score, are calculated to ensure the model performs well in real-world applications. If the validation results are unsatisfactory, the model can be returned to the model training department for readjustment of the training strategy or the number of iterations.

[0052] The implementation process of the model training and optimization module is as follows: First, a data input unit is established to input multimodal data into the model training unit. Then, the model is trained by selecting an appropriate loss function and optimization algorithm. Finally, the model is verified for accuracy and evaluated for performance in the validation and testing unit to ensure that the model has good practical application capabilities.

[0053] The assessment and prediction module, which is connected to the model training and optimization module, uses the optimized model data to estimate the students' cognitive state in real time, and generates the future cognitive trajectory based on the model's prediction results, thus generating the assessment results. Specifically, in this embodiment, the assessment and prediction module utilizes the output of a dynamic thought cognition model, combined with advanced prediction algorithms, to assess students' thought states and predict their future performance.

[0054] The evaluation and prediction module includes a data receiving unit, an evaluation calculation unit, a prediction model unit, and a result output unit. These components are connected via data transmission, forming a complete evaluation and prediction chain.

[0055] The data receiving department receives data on students' ideological states at different times from the dynamic ideological cognition model module. and observed values These processed data provide foundational information for subsequent assessments and predictions, ensuring the accuracy and completeness of the input data.

[0056] Next, the evaluation and calculation department conducts a comprehensive assessment based on the received thought state and observation data. This process uses a weighted scoring model to comprehensively consider the student's various performance aspects; the calculation formula is as follows: ; In the formula, Represents an individual In time The overall evaluation score, The first one is related to the state of mind. Items, This indicates the weight of the corresponding indicator. This indicates the number of indicators considered in the comprehensive evaluation.

[0057] The output of the evaluation calculation unit is a comprehensive evaluation score, which provides important input for subsequent prediction models. This score then enters the prediction model unit, where machine learning algorithms (such as support vector machines, random forests, or deep learning models) are used to predict future performance. The prediction model is expressed as follows: ; in, For individuals In the future Expected performance during the period For the established prediction model, These are additional feature data (such as learning environment, teacher feedback, etc.). The model's task is to predict an individual's future performance based on current assessment scores and contextual features.

[0058] Finally, the results output department is responsible for presenting the evaluation and prediction results to the user in a visual manner. The output includes the individual's overall evaluation score. Predicted future performance And corresponding confidence intervals, so that educators can accurately understand and predict assessment results, and then develop more effective educational intervention programs.

[0059] The attribution analysis module, which is connected to the assessment and prediction module and the model training and optimization module, uses the assessment results and optimized model data to quantitatively analyze the impact of different educational activities on changes in students' cognitive state and generate an attribution report with visualization effects. Specifically, in this embodiment, the attribution analysis module analyzes the impact of different educational activities on changes in students' mental state and generates attribution reports that educators can refer to.

[0060] The attribution analysis module includes a data acquisition unit, an indicator definition unit, a statistical analysis unit, and a report generation unit. These components are connected through a data flow to form a complete attribution analysis process.

[0061] First, in the data acquisition department, assessment results and optimization model data are obtained from the assessment and prediction module. This data includes changes in students' mental state, assessment scores, and corresponding external intervention measures, ensuring that the basic information for subsequent analysis is complete and accurate.

[0062] Secondly, in the indicator definition section, multiple quantitative indicators are defined for different educational activities to quantify changes in students' mental states. These quantitative indicators may include variables such as homework completion rate, classroom participation, online learning time, and emotional feedback. These indicators are normalized using the following formula for easier comparison: ; In the formula, Represents an individual In educational activities Standardized scoring on Represents an individual In educational activities The original score on This indicates the educational activity The set of scores for all individuals in the set. and These represent educational activities. Middle score set The minimum and maximum values.

[0063] Subsequently, in the Statistical Analysis Department, statistical analysis methods (such as regression analysis) were used to analyze the quantitative indicators to determine the impact weight of each educational activity. A multiple regression model was used to calculate the contribution of each educational activity to changes in students' mental state; the model formula is expressed as follows: ; In the formula, Represents an individual Changes in mindset Represents the intercept term. Indicates the first The weighting coefficient of each educational activity Indicates the error term. Indicates all The educational activity is for the first The sum of linear contributions from changes in an individual's state of mind.

[0064] When conducting regression analysis, the attribution analysis module also incorporates an external intervention matrix to ensure the quantification of the direct and indirect impacts of each educational activity on changes in students' mental states. External Intervention Matrix This matrix is ​​used to describe the relationship between different educational activities and student status, with each element representing the degree of influence of each activity on the student.

[0065] Finally, in the report generation department, an attribution report is generated based on the above analysis results. This report clearly demonstrates the specific impact of each educational activity on changes in students' cognitive states, presented to educators in a visual format. The report includes the weight of each educational activity, impact assessment, and recommended measures to help educators better understand the effectiveness of their educational activities.

[0066] The intelligent intervention path recommendation module, which is connected to the assessment and prediction module and the attribution analysis module, plans the optimal educational intervention path using a cost function based on the student's ideological goals and assessment results. Specifically, the intelligent intervention path recommendation module in this embodiment provides educators with personalized intervention path recommendations to optimize students' mental state and learning outcomes.

[0067] The intelligent intervention path recommendation module includes a data receiving unit, a target state setting unit, a path calculation unit, and a result output unit. These components are tightly connected through data flow, forming a complete intervention path recommendation process.

[0068] First, in the data receiving section, the module receives evaluation results and attribution reports from the evaluation and prediction module. This data includes the student's current mental state vector and influencing factors, which are crucial for subsequent analysis, ensuring the accuracy and validity of the basic information.

[0069] Next, the target state setting department sets the target ideological state based on the evaluation results. Furthermore, the key factors influencing the state of mind are analyzed. The target state of mind can be based on historical data or the educator's expected settings, and can be represented as: ; In the formula, This represents the vector of the ideal thought state to be achieved, with each component... Ensure that the educational goals are specific.

[0070] Subsequently, in the path calculation department, the optimal intervention path from the current thought state to the target thought state is calculated using a cost function based on an optimization algorithm. The cost function calculation formula is: ; In the formula, Represents the total cost of the entire path, and represents the cost from the current time. Until the end time Total loss, Indicates each time step The costs are accumulated over time. Start to Time Finish, Represents an individual In time The hidden thought state vector, It is the target thought state vector. Denotes the weighted square norm. Indicates time The applied intervention vector, The weighted square norm representing the intensity of the intervention. This indicates the degree of sensitivity to control costs.

[0071] By introducing appropriate optimization algorithms (such as gradient descent, particle swarm optimization, etc.), the intervention measures can be dynamically adjusted. To minimize the cost function Search for the optimal intervention path.

[0072] Finally, in the results output section, the optimized intervention pathway and corresponding implementation recommendations are output as a report. This report includes recommended interventions, implementation steps, and expected outcomes to support educators in developing personalized teaching plans.

[0073] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent assessment system for ideological education based on artificial intelligence data fusion, characterized in that: include: The data acquisition and processing module is used to collect students' online behavior data, offline performance data and subjective feedback data from multiple data sources, and to clean, fuse, label and time-series align the data to generate multimodal student data. The dynamic thought and cognition model module is connected to the data acquisition and processing module. Through the processed modal student data, it adopts a dynamic state-space model and uses state transition equations and observation equations to describe the evolution of students' thought and cognition states, thereby generating model data. The model training and optimization module is connected to the dynamic thought cognition model module. By acquiring the model data in this module, it trains and optimizes the model parameters to generate optimized model data. The assessment and prediction module, which is connected to the model training and optimization module, uses the optimized model data to estimate the students' cognitive state in real time, and generates the future cognitive trajectory based on the model's prediction results, thus generating the assessment results. The attribution analysis module, which is connected to the assessment and prediction module and the model training and optimization module, uses assessment results and optimized model data to quantitatively analyze the impact of different educational activities on changes in students' cognitive state and generate attribution reports with visualization effects. The intelligent intervention path recommendation module, which is connected to the assessment and prediction module and the attribution analysis module, plans the optimal educational intervention path based on the student's ideological goals and assessment results using a cost function.

2. The intelligent assessment system for ideological education based on artificial intelligence data fusion according to claim 1, characterized in that, The data acquisition and processing module specifically includes: Online learning data automatically collected by the learning management system includes course access records, assignment submission status, and forum interaction data; The acquired text data is transformed into a high-dimensional semantic vector representation using a pre-trained language model to facilitate subsequent analysis.

3. The intelligent assessment system for ideological education based on artificial intelligence data fusion according to claim 1, characterized in that, The state transition equation of the dynamic thought cognition model module is: ； In the formula, Represents an individual In time The hidden thought state vector, Represents a nonlinear function. Represents an individual In time The state transition matrix, This represents the hidden state vector from the previous time step. Indicates the input control matrix. Represents the external intervention vector. Indicates process noise; And the observation equation is: ； In the formula, Represents an individual In time Observational data, Indicates parameters The nonlinear observation function, Represents the observation matrix. This indicates observation noise.

4. The intelligent assessment system for ideological education based on artificial intelligence data fusion according to claim 1, characterized in that, The model training and optimization module uses a combination of variational autoencoder and expectation-maximization algorithm to train the optimized model data in order to maximize the accuracy of the model.

5. The intelligent assessment system for ideological education based on artificial intelligence data fusion according to claim 1, characterized in that, The assessment and prediction module includes the following steps: Receive model data and optimize model data; The extended Kalman filter algorithm is used to estimate the students' cognitive state in real time, and the state estimate is updated at each time point. Based on the current state estimate and external intervention information, the trajectory of thought state over a future period is generated through model prediction and calculation. The generated thought state trajectory and its changing trend are output as the evaluation result.

6. The intelligent assessment system for ideological education based on artificial intelligence data fusion according to claim 1, characterized in that, The assessment and prediction module uses multiple assessment indicators to comprehensively analyze the assessment results and generate personalized assessment reports for different student groups, including the development trend of their ideological state and the gap with the target ideological state, so as to help educators formulate corresponding educational strategies.

7. The intelligent assessment system for ideological education based on artificial intelligence data fusion according to claim 1, characterized in that, The attribution analysis module includes the following steps: Obtain evaluation results and optimize model data; Multiple quantitative indicators are defined for each educational activity to quantify changes in students' mental state; Statistical analysis methods were used to analyze the quantitative indicators and determine the impact weight of each educational activity. Generate attribution reports to demonstrate the specific impact of each educational activity on changes in students' cognitive state, for educators' reference.

8. The intelligent assessment system for ideological education based on artificial intelligence data fusion according to claim 1, characterized in that, The attribution analysis module uses external intervention matrices and evaluation results to perform regression analysis to quantify the direct and indirect impacts of different educational activities on changes in students' mental state, and generates a visual report showing the effectiveness analysis of each intervention measure.

9. The intelligent assessment system for ideological education based on artificial intelligence data fusion according to claim 1, characterized in that, The intelligent intervention path recommendation module includes the following steps: Receive the assessment results and attribution reports; Based on the assessment results, set target mental states and analyze the key factors affecting mental states; The optimal intervention path from the current state of mind to the target state of mind is calculated using a cost function based on an optimization algorithm; It provides recommended interventions and corresponding implementation suggestions to support educators in developing personalized teaching plans.

10. The intelligent assessment system for ideological education based on artificial intelligence data fusion according to claim 1, characterized in that, The intelligent intervention path recommendation module uses a cost function for optimization, which includes: ； In the formula, This represents the total cost of the entire path, starting from the current time. Until the end time Total loss, Indicates each time step The costs are accumulated over time. Start to Time Finish, Represents an individual In time The hidden thought state vector, It is the target thought state vector. Denotes the weighted square norm. Indicates time The applied intervention vector, The weighted square norm representing the intensity of the intervention. This indicates the degree of sensitivity to control costs.

Images